Welcome to Day 2 of Citi Global TMT Conference. My name is Atif Malik. I cover U.S. semiconductors, semiconductor equipment, and networking equipment stocks here at Citi. It's my pleasure to welcome Steve Kelley, President and CEO, and Paul Oldham, EVP and CFO from Advanced Energy Industries. We also have Edwin Mok, VP of Strategic Marketing and Investor Relations in our audience. Welcome, guys. Thanks, Atif. Good to be here. Before we begin, Paul has a few comments regarding REG FD. Right. First, thanks for having us here, Atif, and for everyone for joining us today. Just a reminder that any statements we make today may be subject to a number of risk factors. We would encourage you to have a look at our SEC filings for further discussion of those risks. Also, we had our earnings release on August 5, and we do not intend to make any comments on guidance or any updates to guidance at this meeting. Great. Steve, it wasn't too long ago that you reported your Q2 results. Stock has done well this year. Can you start us off with a quick summary of the quarter? Yeah, the second quarter was good for us. We came in ahead of our guidance, and we benefited from a lot of strength in the AI data center market. We were up almost 50% sequentially in data center, so we're off to a good start this year. We also talked about the fact that we think we can sustain this level of data center demand through the end of this year, if not exceed it. We talked about semiconductor. We had a couple of new wins for EVOS and Everest platforms in the conductor etch and deposition applications. We also talked about industrial medical in the call. After seven quarters of correction, we finally showed some growth in Q2, and we think that's the beginning of a trend where we're going to see sequential growth Q3, Q4, and into 2026. Great. For the total company revenue, you mentioned 17% growth for this year. Can you break that down for each of your markets? Yeah, so we said during the call that we would grow 17% year on year. Now, a large portion of that growth will come from data center. Year on year, the data center business will grow 80%. That's being driven primarily by AI data center buildouts. We're also seeing mid-single-digit growth for semiconductor. Industrial medical will decline for the year despite the fact that we saw a turn in Q2. Telecom and networking is relatively flat. What drove the revision on the semiconductor revenue outlook? Yeah, we work closely with our customers, and we just saw some reduction in their forecast for the second half this year. Nothing dramatic. I think it's important to take a step back. If you take a look at our revenue expectation for 2025, it's our second best year ever. The best year was 2022. That was a crazy year. That was the COVID recovery year. Despite the fact we've taken a little step back for the second half, 2025 is going to go down as a very good year in semiconductor for the company. I remember a couple of years ago you launched EVOS and Everest products at Semicon West. Revenue from your next-generation plasma power products is expected to double this year and reach over $10 million. Can you talk about where you see the momentum, whether it is for 2-nanometer or other leading-edge logic and memory applications? Yeah, so we launched these EVOS, Everest, and NAVX products about two years ago. The pull on those products from our customer base has been really impressive. The revenue from some of those early wins in conductor etch in particular is starting to show up in our revenue line this half. That's why we said this year our new product revenue coming from those three products would be between $10 million and $20 million. Now we're going to see that accelerate next year. In 2027, we'll see further acceleration as dielectric etch revenue starts to hit our revenue line as well. We think that these three products, EVOS, Everest, and NAVX, are really the foundation for gaining market share for the rest of this decade. Just to answer your first question, right now they're being evaluated across all of the leading-edge processes, both in logic and in memory. These are very challenging processes for our customers. Some of the structures they're trying to build are very difficult. They're trying to build very deep holes into the silicon structure, and they need new technology. That's why there's been such a pull on these EVOS and Everest products, because they're a considerable step ahead of the existing technology. Steve, just to level set for the audience's sake, if you can just talk about where you're strong, whether it's dielectric or metal side of etch, and where a competitor is stronger. Yeah, historically, Advanced Energy has been very strong in what's called conductor etch, and we've got most of the market share there. The other type of etch, called dielectric etch, we've not been strong in, so we have a very small position there. The opportunity for these new products is for us to establish a significant beachhead in dielectric and then to expand our conductor etch share. In addition to the design wins on leading-edge processes, I think there's a second step where, in some cases, our customers may be able to use our technology to upgrade existing tools. That's an upside opportunity, not yet confirmed, but certainly a possibility. It remains a duopoly type of market. It's a good market for both you guys, right? Yeah, I would say there's two major players in dielectric, Daihen Corporation in Japan and MKS Instruments in the U.S. I know when my clients are calling me to kind of explain to them what these power supplies are, and they're thinking of something like car battery or something, and I'm trying to tell them these are not, this is super complicated RF match and all this stuff. Can you talk about the competitive advantage that you have on these new products and the opportunities in dielectric? Maybe instead of going into too much technical detail, I could talk about the benefits of our products. The issue our customers have with these leading-edge processes is the existing technology results in worse yield and lower throughput. The two most important metrics in a wafer fab are throughput, wafer throughput, and wafer yields. That's why these new products that we have introduced, Everest, EVOS, and NAVX, allow the customer greater control over how the energy is pulsed into the plasma chamber. They could use traditional RF generation, or they could use pulsed DC. We have both of the major mechanisms for delivering power into a plasma chamber, and we think they're both best in class. Before we move on to data center compute, I'm just curious in terms of your outlook into next year, if you have any visibility. I understand you guys are suppliers to your OEMs, and it's difficult to see next year, but any thought of early predictions you have on the semiconductor market next year? Yeah, we're bullish on the market, on the data center market, primarily because of the forecast that we're receiving from our customers. Also, because of the announcements from the hyperscalers about their investment plans for next year, which are pretty robust. We think the investment plans, our forecasts, the fact that power consumption continues to increase, are all factors that work in our favor in data center. On the data center compute side, obviously AI has been the biggest growth driver in the market. Can you talk about where you participate there, who are you working with, and your dollar content opportunity? Yeah, this is a sensitive area because our business is really customized solutions for hyperscale applications. We don't talk publicly about exactly what those solutions are. We work closely. One of the things that's changed in the data center market is with the advent of AI data centers, the cadence of new solutions has sped up tremendously because we're seeing solutions change every year now. That's tied pretty closely to the introduction of NVIDIA GPUs. What that has done is it's drawn us closer together with our customers because they need to work with us from the beginning of the design so that we can have the power solution ready when they have the rest of the system solution ready. We've gone very deep with a select group of customers, and we have a number of projects with each of our target customers. The issue, really, the only constraint we have is engineering bandwidth. We have to be careful that we don't get involved in projects that aren't going to pay off. We've done a pretty good job of it. Again, just for the sake of the audience, are there OpEx synergies or revenue synergies between the power supplies you're doing on the semiconductor side and what you're doing on the data center side? I'll talk about the revenue, and you can talk about the OpEx. Sure. I think one of our advantages as a company is that we are focused on precision power. Think about that as the high end of the power delivery market, essentially. We develop technology blocks for various applications, and we do that in a modular way. That means we have hardware modules, software modules. That allows us to reuse technology across the company. Today we use technology developed in hyperscale in some of our latest and greatest semiconductor products, and vice versa. We have semiconductor cooling technology, liquid cooling technology that we can deploy for data center when we reach that point sometime in the next couple of years. This is a big advantage. We have more technologies under one roof than I think any of our competitors. Moving forward, that's a way to boost our efficiency as well as improve the development speed of new products. Yeah, I think those efficiencies of using these technology blocks across the enterprise save us time in development. It also gives us, you know, arguably best-in-class capability as we're able to leverage these different parts. In some cases, we insource to ourselves, so we don't have to buy it from a third party for some of these building blocks. That increases our speed to time to market as well as lowers our cost. Paul, how much incremental CapEx are you spending this year related to data center, and how much do you plan to spend for next year? Historically, we spend somewhere between 2% and 4% of revenue in CapEx. We increased that a little bit about a year ago, and we talked at our analyst day, and then we increased it again at the end of the first quarter as we saw this building demand for data center products. Today we expect to spend somewhere in the neighborhood of 6% or a little higher of revenue of CapEx. That will support our ability to improve both our capability, our infrastructure to support this much higher power products that are going through our factories, as well as capacity, just actual manufacturing equipment as we go through the various steps of manufacturing. We've been able to grow this business a lot. It's more than double in Q2 than it was a year ago and up 50% from Q1. That's been tough to stretch that much. I think getting this capacity in place will allow us to operate at a more normalized rate, actually lower our costs a little bit from a ramp and cost-to-quality perspective, and give us optionality and ability to capture upside in the future. The mix of GPU versus ASIC customers, I know it's sensitive, but the power profile is quite different for GPUs, a lot more power than ASIC. What's your mix for? Yeah, we don't have any insight to that. The customer doesn't need to disclose their design to us. What we do is we take their spec, some of the inputs, some of the outputs, and then we work around that spec. We don't know exactly the mix between NVIDIA GPUs and ASIC. Let's talk about the industrial and the medical market. You talked about industrial and medical bottoming in Q1, and we did see some sequential improvement in Q2, continuing into the second half. What's the pace and the scale of recovery, and is there any demand pull-in? Yeah, we haven't seen any demand pull-in for tariffs. We ask the question every month and it just really hasn't appeared on our radar screen. I get asked the question, why are we convinced that it's a recovery? A couple of reasons. One is we sell half of these industrial medical products through the distribution channel, and we track the channel metrics every month. For five quarters, the inventory has come down. We've seen resales either go up or stabilize for the last few quarters, and we see basically just the inventory is getting pretty close to quarterly demand. We think in that channel, we're close to a healthy balance. It's a return to equilibrium. The other data points are anecdotal, but in my discussions with industrial medical customers, more of them are telling me they've worked their way through the inventory issues. A lot of industrial medical customers got stuck with a lot of inventory at the end of the COVID supply chain crisis, and it's taken them the better part of two years to work their way through it. That's good news. It means the demand for our new products will be returning. The third data point for me is the amount of design wins we've earned in industrial medical the past three years. Not all those will go to production, but now that some of the inventory is getting cleared out, many of those will go to production. I think we've started to gain market share in industrial medical next year. What is the underlying driver for replacement for your systems? The data center is clear as the Rubin and the next platform that's going to drive the powerful profile. Is it more of the replacement cycle that drives the steady demand in this market? Yeah, industrial medical is an interesting market because it's very sticky, and the life cycle is typically very long. Once you're in, you're in. Some of these applications continue to buy customized power supplies for 10years- 20 years. It's difficult to displace incumbents. That's great if you're an incumbent, not so great if you're trying to displace somebody else. Our wins are on new products, new applications, new products our customers have developed. In addition to the organic growth, to speed things up in industrial medical, we would intend to make acquisitions in that area to build critical mass. Which end markets are stronger within INM and which are weaker? Yeah, you know, we're seeing strength, obviously, in industrial. That area is booming. We're seeing strength in test and measurement, in factory automation. Those are the three that spike out right now. Right. Let me Let me pause there to see if there are any questions in the audience. If you have a question, please raise your hand. Thanks, Atif. I just thought if we could talk a little bit more about the semi-equipment business. You talked about new products that start to really help a little bit this year, much more next year, year after. Can you talk about your broader outlook for semis for next year? How do you think about that market, what's intake? Yeah, what we said about the semi market for next year is we expect a flattish market. If you dive a little deeper, you know, we expect to see continued weakness for trailing edge processes. That means China as well as the Western trailing edge demand, offset by leading edge demand. That's going to be driven by continued demand from AI applications. Okay. In In data center, can you talk about, like you said, hyperscalers, can you talk about your engagement with enterprise customers? Because there is like this talk about broadening of AI. Yeah, our primary focus is on hyperscalers because that's where the volume is. I mentioned in the earnings call a few weeks ago that we're also seeing interest from other customers, which would include enterprise customers. The way we're addressing those opportunities is to see how much existing technology we could reuse. If we can do something that doesn't drain engineering resource to any significant degree, then we go ahead and do it. We're not going to deprioritize the hyperscale requirements to go chase enterprise opportunities. It's a fine balance. Atif Braun. Hardest place to get to, sorry. Just wanted to hear your comments on kind of ability to scale in the data center. I think you alluded to the fact that you're kind of gated by engineering capacity and therefore you want to prioritize the right places. Let's say that this growth continues. How confident do you feel in your ability to meet demand and then, you know, sustainably over a long period of time get new talent into the market to serve what's a pretty big opportunity? Yeah, that's a good question. There are two elements to that. One is actual production capacity, and we're making substantial investments. We'll continue to make substantial investments to stay ahead of our customers on their demand. The second is human capacity. You know, it's development capacity. At Advanced Energy Industries, we have roughly 1,400 design and development engineers. We have a lot of people, and a significant number of those are working on data center applications. We have the ability to shift talent around, which we do, and we've moved more people into that group. At the end of the day, the other thing we are doing is expanding our footprint from a development standpoint. Most of our engineers are located in Asia, in Hong Kong, Taiwan, Philippines, but also in the U.S., in Minneapolis, on the West Coast. We've established a center on the East Coast in the Massachusetts area where we've added 50 engineers in the past two years. We're moving to where the engineers are. I think there's always a limit on how much you can do, and the key for us moving forward is making sure that we get the best return on investment for our engineering team. Paul, on the model, the tariff impact to 2Q gross margins was around 100 basis points. Given the increased tariff rates in the Philippines and Malaysia, do you expect a greater impact in the second half? What's kind of baked into your guidance? Yeah, it's a good question. Obviously, tariff rates increased from roughly 10% globally to higher. Most Southeast Asian countries are now around 19%, including in Malaysia and in the Philippines where our factories are. At the same time, we've been able to mitigate some of the costs we incurred in Q1. Roughly, we think those kind of offset each other, the increased impact of tariffs versus actions we've been able to take so far to mitigate. We expect tariffs to continue, we said, at this rate or a little higher. That's embedded in our guidance. Even with that tariff headwind, we expect gross margins in the third quarter to improve a little bit to sort of mid 38%. We expect the fourth quarter to be between 39% and 40%. That improvement is largely driven by the closure of our factory in China and continued consolidation of our manufacturing footprint. If you exclude tariffs, we would be running over 40% by the time we exit the year, which was our goal. We think we're, in terms of what we can control, on track or a little bit ahead in terms of improving gross margins in the near term. As we look out beyond the near term, we think there's other levers that we have that can offset the impact of tariffs. Those include continued optimization and efficiency improvements in manufacturing, driving material costs lower. Obviously, as our new products start to come in and take a bigger impact on the mix, those things can all help us overcome the tariff impact, assuming it sort of stays in this range. With the closure of the China factory, you guys have been on this kind of consolidation timeline for the last couple of years. Can you just help us understand where you are, how much more consolidation is needed across your manufacturing sites? Yeah, so this last quarter in June, we finished production in our last China manufacturing site. That means we closed three sites in China in the last four years. That effort was obviously put in place a while ago to manage geopolitical risk. Of course, it turns out that that seems to be the right thing to have done. In addition, our customers wanted us, for the most part, out of China. We're fortunate that that's been done. In parallel, we've been on a path to do a rooftop consolidation and to consolidate from 15 factories essentially down to about five factories. We started with the biggest ones first, where they had the biggest financial impact, and we're well through that process. As we go into 2026, there's still a number of smaller ones that we'll have to do a little bit of cleanup on. We're excited because this gives us a much more streamlined footprint, gives us the ability to improve efficiency, improve quality, common processes across our factories. Okay. How does the success in the data center business impact your gross margins? What's kind of baked in that 2030 targets you shared at the 2024 analyst day? Yeah, it's a good question. If you look at our margin profile, we've been pretty open that our data center margins are below the corporate average. The good news is we've focused on our strategy to play more in the customized part of this market, the higher power that plays to our strengths. We've been able to close that gap. Today, the gross margins we've said are approaching or near corporate average. They're not nearly as much of a drag as they used to be. In fact, the way we think of it is changing the mix could have a plus or minus 50 basis point impact. I think a proof point of that is in the first quarter, we had a pretty heavy mix of semiconductor products. In the second quarter, that shifted and we had a much heavier proportion of data center products. In fact, I think it was close to a third, our highest ever, and yet margins actually improved a little bit sequentially. We've been able to weather this change in mix because we've been able to make pretty significant improvements in the underlying margins of those products. All right. On the capital allocation, what are the priorities and what's the strategy on M&A? I will cover the priorities and Steve can talk about our approach to M&A. Clearly, we think the best use of capital is to continue to grow the company. Obviously, we're investing some of that into growing the company. Our CapEx, as I mentioned, is going to be higher for the next several quarters. We're going to continue to invest in engineering and other things. We think we can grow the company through smart M&A where we can add essentially scope to what we sell and then leverage the scale of the company. That will consume, we think, the majority of our capital. We also have an opportunistic share repurchase plan in place. From time to time, as we see dislocations in the market, we're able to be pretty aggressive on that plan. In fact, in the first quarter or early in the second quarter, we actually repurchased over $20 million of stock over the course of a couple of weeks to take advantage of the market situation we saw. We bought that stock back at a little over $83 per share. In hindsight, that's a pretty good investment and we're able to get quite a few shares back for that. Yeah, so our M&A strategy, you know, we think we can grow pretty well organically in semiconductor and data center. We think we've got the ingredients in-house. The area where we need to bulk up is in industrial medical. I mentioned it earlier in this presentation, but we think there's a partial roll-up strategy that will help build critical mass in industrial medical and accelerate our share gain program. That's where the primary focus is. The secondary focus is on technology tuck-ins, like we did last year with a company called Eridi. That was a big success for the company because we were able to deploy the Eridi technology immediately into some of our more advanced platforms. We're actively looking for acquisition opportunities. All right. In the end market, we didn't touch on the telecom and networking where we have seen a recovery in terms of spending from the service providers. I was curious if you have also participated in that recovery and what's the outlook? You know, it's an interesting market for us. It's just not big enough for us to dedicate a lot of design resource. Again, we shifted over to hyperscale and to a lesser extent industrial medical. We're treating the telecom and networking market opportunistically, which means if we could put some technology blocks together to satisfy the customer need, then we'll do it. We're not going to peel off engineering teams away from hyperscale or industrial medical. Great. We have a few minutes left. If anyone has questions, you can come up and talk to the management team. Thank you, Steve and Paul, for coming to the Tech Conference. Thanks, Atif. Thank you, Atif. Yeah, it's everyone.